Cannot save TFTapasModel as SavedModel

[ahmedlone127]

System Info

• transformers version: 4.20.1
• Platform: Linux-5.4.188+-x86_64-with-Ubuntu-18.04-bionic
• Python version: 3.7.13
• Huggingface_hub version: 0.8.1
• PyTorch version (GPU?): 1.12.0+cu113 (True)
• Tensorflow version (GPU?): 2.8.2 (True)
• Flax version (CPU?/GPU?/TPU?): not installed (NA)
• Jax version: not installed
• JaxLib version: not installed
• Using GPU in script?: No
• Using distributed or parallel set-up in script?: No

Who can help?

@Rocketknight1 @gante

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

from transformers import TapasTokenizer, TFTapasModel
import pandas as pd

tokenizer = TapasTokenizer.from_pretrained("google/tapas-base")
model = TFTapasModel.from_pretrained("google/tapas-base")

data = {
    "Actors": ["Brad Pitt", "Leonardo Di Caprio", "George Clooney"],
    "Age": ["56", "45", "59"],
    "Number of movies": ["87", "53", "69"],
}
table = pd.DataFrame.from_dict(data)
queries = ["How many movies has George Clooney played in?", "How old is Brad Pitt?"]

inputs = tokenizer(table=table, queries=queries, padding="max_length", return_tensors="tf")
outputs = model(**inputs)

last_hidden_states = outputs.last_hidden_state

model.save_pretrained("test",saved_model=True)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
[<ipython-input-11-637c488e6341>](https://localhost:8080/#) in <module>()
----> 1 model.save_pretrained("test",saved_model=True)

2 frames
[/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py](https://localhost:8080/#) in autograph_handler(*args, **kwargs)
   1145           except Exception as e:  # pylint:disable=broad-except
   1146             if hasattr(e, "ag_error_metadata"):
-> 1147               raise e.ag_error_metadata.to_exception(e)
   1148             else:
   1149               raise

ValueError: in user code:

    File "/usr/local/lib/python3.7/dist-packages/transformers/modeling_tf_utils.py", line 806, in serving  *
        output = self.call(inputs)
    File "/usr/local/lib/python3.7/dist-packages/transformers/modeling_tf_utils.py", line 981, in run_call_with_unpacked_inputs  *
        return func(self, **unpacked_inputs)
    File "/usr/local/lib/python3.7/dist-packages/transformers/models/tapas/modeling_tf_tapas.py", line 1008, in call  *
        outputs = self.tapas(
    File "/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py", line 67, in error_handler  **
        raise e.with_traceback(filtered_tb) from None

    ValueError: Exception encountered when calling layer "tapas" (type TFTapasMainLayer).
    
    in user code:
    
        File "/usr/local/lib/python3.7/dist-packages/transformers/modeling_tf_utils.py", line 981, in run_call_with_unpacked_inputs  *
            return func(self, **unpacked_inputs)
        File "/usr/local/lib/python3.7/dist-packages/transformers/models/tapas/modeling_tf_tapas.py", line 790, in call  *
            embedding_output = self.embeddings(
        File "/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py", line 67, in error_handler  **
            raise e.with_traceback(filtered_tb) from None
    
        ValueError: Exception encountered when calling layer "embeddings" (type TFTapasEmbeddings).
        
        in user code:
        
            File "/usr/local/lib/python3.7/dist-packages/transformers/models/tapas/modeling_tf_tapas.py", line 223, in call  *
                col_index = IndexMap(token_type_ids[:, :, 1], self.type_vocab_sizes[1], batch_dims=1)
        
            ValueError: Index out of range using input dim 2; input has only 2 dims for '{{node tapas/embeddings/strided_slice_2}} = StridedSlice[Index=DT_INT32, T=DT_INT32, begin_mask=3, ellipsis_mask=0, end_mask=3, new_axis_mask=0, shrink_axis_mask=4](token_type_ids, tapas/embeddings/strided_slice_2/stack, tapas/embeddings/strided_slice_2/stack_1, tapas/embeddings/strided_slice_2/stack_2)' with input shapes: [?,?], [3], [3], [3] and with computed input tensors: input[3] = <1 1 1>.
        
        
        Call arguments received:
          • input_ids=tf.Tensor(shape=(None, None), dtype=int32)
          • position_ids=None
          • token_type_ids=tf.Tensor(shape=(None, None), dtype=int32)
          • inputs_embeds=None
          • training=False
    
    
    Call arguments received:
      • self=tf.Tensor(shape=(None, None), dtype=int32)
      • input_ids=None
      • attention_mask=tf.Tensor(shape=(None, None), dtype=int32)
      • token_type_ids=tf.Tensor(shape=(None, None), dtype=int32)
      • position_ids=None
      • head_mask=None
      • inputs_embeds=None
      • output_attentions=False
      • output_hidden_states=False
      • return_dict=True
      • training=False

Expected behavior

It is supposed to make a SavedModel but instead, I get this error mentioned above. The SavedModel is needed for TensorFlow Serving .

[amyeroberts]

Following merging of #18153 the reproduction snippet runs on main without error.